lable at ScienceDirect

Quaternary Science Reviews 166 (2017) 363e379

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

Controls on dryland mountain landscape development along the NWSaharan desert margin: Insights from Quaternary river terracesequences (Dad�es River, south-central High Atlas, Morocco)

M. Stokes a, *, A.E. Mather a, M. Belfoul b, F. Faik b, S. Bouzid b, M.R. Geach a, c, f, P.P. Cunha d,S.J. Boulton a, C. Thiel e, f

a School of Geography, Earth and Environmental Sciences, Plymouth University, Drake Circus, Plymouth, Devon, PL4 8AA, UKb Structural Geology and Thematic Mapping Laboratory, Earth Sciences Department, Faculty of Science, Ibn Zohr University, PO Box 8106, Hay Dakhla,Agadir, 80000, Moroccoc ATKINS, Woodcote Grove, Ashley Road, Epsom, KT18 5BW, UKd Marine and Environmental Sciences Centre, Department of Earth Sciences, Universidade de Coimbra, Coimbra, Portugale Leibniz-Institut für Angewandte Geophysik, Stilleweg 2, 30655, Hannover, Germanyf Nordic Laboratory for Luminescence Dating, Aarhus University, and Centre for Nuclear Technologies, Technical University of Denmark (DTU), Risø Campus,Denmark

a r t i c l e i n f o

Article history:Received 23 December 2016Received in revised form13 April 2017Accepted 23 April 2017Available online 8 May 2017

Keywords:PleistoceneClimate dynamicsPaleogeographyAfricaFluvial geomorphologyOptical methods

* Corresponding author.E-mail address: mstokes@plymouth.ac.uk (M. Stok

http://dx.doi.org/10.1016/j.quascirev.2017.04.0170277-3791/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

This study documents river terraces from upstream reaches of the Dad�es River, a major fluvial systemdraining the south-central High Atlas Mountains. Terraces occur as straths with bedrock bases positionedat 10 m altitudinal intervals up to 40 m (T1-T5) above the valley floor, becoming less common between50 and 140 m. The rock strength, stratigraphy and structure of the mountain belt influences terracedistribution. Terraces are absent in river gorges of structurally thickened limestone; whilst well-developed, laterally continuous terraces (T1-T4) form along wide valleys occupying syncline structuresdominated by weaker interbedded limestone-mudstone. Terrace staircases develop in confined canyonsassociated with weaker lithologies and influence from structural dip and stratigraphic configuration.

Terraces comprise a bedrock erosion surface overlain by fluvial conglomerates, rare overbank sandsand colluvium. This sequence with some OSL/IRSL age control, suggests terrace formation over a 100 kaclimate cycle with valley floor aggradation during full glacials and incision during glacial-interglacialtransitions. This integrates with other archives (e.g. lakes, glaciers, dunes), appearing typical of land-scape development along the NW Saharan margin south of the High Atlas, and similar to patterns in thewestern-southern Mediterranean. The 100 ka climate cycle relationship suggests that the terracesequence documents Late-Middle Pleistocene landscape development.

Consistent altitudinal spacing of terraces and their distribution throughout the orogen suggests sus-tained base-level lowering linked to uplift-exhumation of the High Atlas. Low incision rates (<0.2 mma�1) and general absence of terrace deformation suggests dominance of isostatically driven base-levellowering with relief generation being Early Pleistocene or older.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

River terraces are bodies of fluvial sediment with a step-likemargin and flat topped surface that grade topographically towardsa valley centre and in a downstream direction (Stokes et al., 2012a).They are commonly preserved along river valleymargins around the

es).

world and are considered to have regional lithostratigraphicimportance (Bridgland and Westaway, 2008a,b; 2014). A terraceforms when a river incises, abandoning the former floodplain andchannel belt (Leopold et al., 1964). Successive periods of fluvialsediment aggradation, river incision and floodplain-channel aban-donment can form inset river terrace staircases (Starkel, 2003). Thefluvial sedimentation is generally a function of climate-relatedsediment supply and flood regime, whilst river incision is deter-mined by base-level lowering, normally driven by combinations oftectonics, climate and eustacy (e.g. Bridgland andWestaway, 2008a;

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379364

Vandenberghe, 2015). Global database compilations of river terraceresearch have highlighted the importance of terraces as archives ofLate Cenozoic (mainly Quaternary) environmental change and theirrelationship to unravelling the development of topographic relief(Bridgland and Westaway, 2008a,b; 2014). Much of this research isfocussed on large, mid latitude northern hemisphere river systemsthat have evolved under fluctuating glacial, periglacial and humidclimatic conditions due to their ice sheet proximity and connectiv-ity. With few exceptions (see Bridgland and Westaway, 2008a andreferences therein), there is a notable research gap from lowerlatitude regions where fluvial systems have evolved in environ-ments with negligible-minimal direct ice sheet influence underdrier continental conditions. One such area is NW Africa and itsinland arid region south of the High Atlas Mountains (Fig. 1).

The NW Saharan Desert margin is bordered by the 3e4 km reliefof the High Atlas Mountains of Morocco. Its topography, and thelower relief ~2 km Anti-Atlas Mountains to the south (Fig. 1), marksthe southern limits of ongoing Alpine tectonic collision betweenAfrica and Europe (Dewey et al., 1989) and is a key area of climateinterchanges between the Atlantic, Mediterranean and continentallow latitude regions of Central-West Africa (Tjallingii et al., 2008).The mountain landscape origins relate to accelerated Plio-Quaternary uplift (Frizon de Lamotte et al., 2000) and globalclimate cooling and fluctuations throughout the Quaternary (Plaziatet al., 2008). Quaternary sediments and landformsoccur throughoutthe region (Plaziat et al., 2008) but river terrace research is limited,being geographically restricted to coastal areas of theAnti-Atlas (e.g.Weisrock et al., 2006) and localised parts of the northern High Atlas(Delcaillau et al., 2016). Coastal river terrace studies are eitherregional treatments, using river-marine terraces for numerical upliftmodelling (e.g. Westaway et al., 2009) or are localised studies highresolution dating studies of a single Late Pleistocene terrace levelsite for detailed climatic insights (e.g. Weisrock et al., 2006). TheseAtlantic draining rivers have a strong climate-eustatic base levelcontrol, have developed in low mountain relief and are influencedfrom humid Atlantic weather. However, these coastal systems arenot typical of inland, where drainage has evolved under

Fig. 1. A) Topography of NW Africa and Quaternary landscape study locations referred to wstudies; C ¼ glacial cirque study; S ¼ Dad�es River study area; L ¼ mountain lake study). B) TeNote River Dad�es study region (black rectangle: Figs. 2 and 3). OB ¼ Ouarzazate Basin; Drsystems. Modified from Michard (1976) and Carte G�eologique du Maroc (1985).

considerably drier continental climatic conditions, greater reliefvariability and absence of eustatic base level control. The southernflanks of the High Atlas provide an excellent area to study riverterraces and their role for unravelling Quaternary fluvial landscapedevelopment typical of the dryland continental interior of NW Af-rica. Here, rivers are 1) sourced from some of the highest mountainrelief, 2) are routed southwards from mountain through to lowerlying desert climatic regions and 3) cross all of the key tectoniccomponents of the High Atlas orogenic system.

In this study we focus on a 60-km-long upstream reach of theDad�esRiver (Fig.1), a principal river systemdraining the south-centralHigh Atlas in the characteristic dryland continental interior of NWAfrica. The terraces are used to explore the interplay of geological,climate and tectonic base level controls on Quaternary fluvial land-scape development. The relationship of the terrace sequence to therock strength, structure and stratigraphic sequence of the High Atlasorogen is assessedasapassivegeological control on terrace formation.Terrace-climate relationships are examined using an OSL/IRSL datedterrace sediment sequence, providing insights into the timing offluvial aggradation and incision patterns and their relationship to a100 ka climate cycle. The terrace sequence is then used to explorespatial and temporal patterns of base-level lowering through incisionrate quantification and its relationship to the timing andmechanismsof tectonic relief generation of the High Atlas orogen.

2. Geological and geomorphological background

2.1. Regional geology and drainage evolution

The Dad�es River is one of the larger drainage systems in NWAfrica forming the principal perennial river that drains the south-central High Atlas Mountains of Morocco (Fig. 1B). On enteringthe Ouarzazate Basin, the Dad�es River flows westwards along thenorthern margin of the Anti-Atlas. At Ouarzazate town, the Dad�esRiver joins the Draa River, turning SE with routing across the Anti-Atlas through the Draa Gorge before turning WSW to join theAtlantic Ocean.

ithin text (D ¼ dune studies; T ¼ terrace studies; G ¼ glacier moraine study; F ¼ fanctonic and geological configuration of Morocco and the south-central High Atlas region.¼ Draa gorge and capture region; NAF/SAF/TAF ¼ Northern/Southern/Tell Atlas Fault

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 365

This drainage configuration relates to the long term tectonicdevelopment of NWAfrica. The High Atlas is an ENE-WSWorientedAlpine orogenic system formed by African-European plate suturing(Dewey et al., 1989; Frizon de Lamotte et al., 2008). Relief genera-tion of 2e4 km has involved the inversion of a Mesozoic intra-continental rift system by regional thrust faulting and folding (e.g.Gomez et al., 2000) and thermal-related isostatic uplift related to amantle plume underlying the High Atlas (e.g. Missenard et al.,2006). The mountain belt displays common characteristics of acollisional orogenic system (Figs.1B and 2): a high relief ‘axial zone’,bordered by thrust front (Northern and Southern Atlas faults [NAF/SAF]) and peripheral foreland basins around the relief margins(Chellai and Perriaux, 1996; El Harfi et al., 2001). The stages andtiming of High Atlas relief generation are debated but generally

Fig. 2. A) Geology of the study area within the High Atlas orogenic system, illustrating key lorogenic cross-section. Map and cross-section modified from Carte G�eologique du Maroc (1

considered to have occurred in two stages during the Early and LateCenozoic (Frizon de Lamotte et al., 2000).

Analysis of the High Atlas drainage pattern by Babault et al.(2012) describes rivers that are longitudinal (strike-orientated) ortransverse to the mountain belt structure. This research suggeststhat early drainage is inherited from growing upper crust fault-foldstructures, with later drainage development configured to ampli-fied N-S regional slopes as topography grows. The modern Dad�esRiver is oblique to the orogen structure, suggesting it is a longerlived longitudinal drainage system (Babault et al., 2012) but onethat contains large scale knick zones linked to more recent Plio-Quaternary uplift and relief generation (Boulton et al., 2014).

At continental scale, the modern Dad�es-Draa system (Fig. 1B) isrelated to capture of the internally drained Ouarzazate Basin by an

ithostratigraphic units and major tectonic and stratigraphic structures. B) North-South975, 1990; 1993). See Table 1 for stratigraphic and lithologic information.

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379366

Anti-Atlas drainage (St€ablein, 1988). The capture is probably aMiddle-Late Pleistocene occurrence based upon fan surface datingfrom sites north of Ouarzazate town (Arboleya et al., 2008).

2.2. Geology and geomorphology of the Dad�es River

The headwaters of the modern Dad�es River lie at altitudes of2800e3400 m (above sea level), with the river entering the Ouar-zazate Basin at ~1550 m (Fig. 3A). The study area (Fig. 3B) encom-passes key geological and morphological components of the HighAtlas orogenic system (Figs. 2 and 3): 1) high relief (3400e2000 m)fold-thrust belt (FTB), 2) intermediate relief (1900e1600 m)wedge-top basin (WTB), 3) intermediate relief (1900e1600 m)thrust front (TF) and 4) low relief (1600e1500 m) foredeep basin(FB) (orogen terminology sensu DeCelles and Giles, 1996).

The FTB forms themid to upstreamparts of the study area (Fig. 3).Jurassic marine limestone and mudstone lithologies are dominant(Carte G�eologique du Maroc, 1990, 1993), corresponding to a Meso-zoic rift system that forms the central-eastern High Atlas (Warme,1988). This bedrock is affected by a series of regional fold and thrustfault structures (Fig. 2). In the upstream FTB the Dad�es River cuts

Fig. 3. A) Dad�es River catchment, relief and Strahler stream ordering. B) Detailed study area(white line) between the towns of Boumalne du Dad�es (BDD) and M’smerir (M), river terr4 ¼ WTB; 5 ¼ TF; 6 ¼ FB) annotated with geology and regional bedrock dip configurationOuidane Fm, O ¼ Ouchbis Fm, JC ¼ Jbel Choucht Fm: See Fig. 2 and Table 1).

through theSE limbof a symmetricopenanticline ina series of deeplyincised (>100 m depth) high sinuosity canyons (Figs. 2, 3 and 4A).Downstream, the folding changes to an asymmetric synclinewith theDad�es River routed southwest along the fold axis in an open valleyconfigured by the syncline (Figs. 2, 3 and 4B). A route deviation to thesoutheast occurs at the Tarhía n' Dad�es gorge (Figs. 2, 3 and 4C)wherethe Dad�es River cuts a short 0.5 km long, 150 m deep and 20e50 mwide route through a Lower Jurassic limestone ridge (Stokes et al.,2008). In the downstream distal part of the FTB, thrust faulting hasstructurally thickened Lower Jurassic limestones causing the Dad�esRiver to cut a deeply incised (>200 m), ~5 km long and 20e100 mwide gorge, the Dad�es Gorge (Figs. 2, 3 and 4D; Stokes et al., 2008).

The Dad�es River emerges from the Dad�es Gorge into the AïtSedratt WTB (Figs. 2, 3 and 4E). Plio-Quaternary terrestrial fan andfluvial conglomerates are organised into a syncline that hasdeveloped between the FTB and the TF (Carte G�eologique duMaroc,1990, 1993). Sediment provenance, palaeocurrents andstratigraphic-structural organisation suggests a complex FTBtransverse and parallel palaeo-drainage routing influenced bythrust faulting and folding (Cavini, 2012), corresponding in part toan ancestral Dad�es River. The modern Dad�es River is routed

showing the drainage divide (black line), the Dad�es River trunk drainage study regionace distribution (red dots) and cross valley profiles (1e3 ¼ proximal, mid, distal FTB;(P-Q ¼ Plio-Quaternary; AO ¼ Aït Ouglif Fm; E-C ¼ Eocene-Cretaceous; BEO ¼ Bin El

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 367

southwest, transverse to the WTB in a relatively open asymmetricvalley flanked by high cliffs on western banks and lower slopes/cliffs on eastern banks (Figs. 3B and 4E).

In the TF region, the Dad�es River passes through a structurallycomplex ENE-WSW striking, southwards verging folded and obliquethrust faulted sequence of Mesozoic-Cenozoic lithologies (CarteG�eologique du Maroc, 1975; Tes�on and Teixell, 2008). The river dis-plays marked orientation changes as it passes through the TF region,showing a dominant configuration to the ENE-WSW strike ofdeformation, with shorter NNW-SSE valley reaches that are

Fig. 4. Bedrock geology and geomorphology study area field imagery: A) High sinuosity bedrdeveloped in mixed-weak strength Lower Jurassic bedrock, mid FTB region. C) Tarhía n’ DadFTB region. E) Open valley form developed in mixed-weak strength Mio-Quaternary bedrockupstream TF region. G) Gorge (arrowed) changing to open valley form (foreground) developevalley form developed in mixed-weak Mio-Quaternary bedrock, FB region. See Table 1 for

transverse to the TF strike (Figs. 2 and 3B). The strike and transverseorientatedvalleys are relativelyopen (Fig. 4F)but short gorge reachesexist where the river routes through Jurassic and Palaeogene lime-stones (Fig. 4G). Cross-sections and stratigraphic-structural analysesby Tes�on andTeixell (2008) suggest that the TF regionhas undergonelow rate (0.3 mm/a�1) tectonic shortening of 7e8 km.

Finally, the Dad�es River emerges from the TF passing south-wards into the (Ouarzazate) FB (Figs. 2, 3 and 4H). Here, Plio-Quaternary terrestrial fan and fluvial conglomerates form thedominant lithology, with some stratigraphic affinity to similar

ock canyon in strong Middle Jurassic bedrock, proximal FTB region. B) Open valley form�es Gorge and D) Main Dad�es Gorge developed in strong Lower Jurassic bedrock, distal, WTB region. F) Open valley form developed in mixed-weak Mio-Quaternary bedrock,d in mixed-weak (Cretaceous) and mixed-strong (Eocene) bedrock, TF region. H) Openstratigraphy, lithology and rock strength characteristics and notation.

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379368

sediments in the WTB (Cavini, 2012). The sediments form a topsurface ~100m above the Dad�es River valley floor (Fig. 3B), with thesurface revealing relict fan-shapedmorphologies fed from drainageoutlets along the TF/FB margin. Further west in central and westernparts of the Ouarzazate Basin, the surface is incised forming astepped landscape with up to 5 inset fan/terrace surfaces (G€orleret al., 1988; St€ablein, 1988; Arboleya et al., 2008). Some higher-older surfaces display folding and fault scarps (S�ebrier et al.,2006; Arboleya et al., 2008) suggesting Quaternary deformation.Seismic activity is infrequent and low magnitude, with historicalshallow earthquakes of magnitude <4.9 distributed throughout theTF/FB margin regions (Medina and Cherkaoui, 1991).

2.3. Quaternary-recent characteristics of the Dad�es River andsurrounding region

The modern Dad�es River is a perennial 5th order trunk drainage(Fig. 3A) fed by ephemeral tributaries (Stokes and Mather, 2015). Themodernchannel occupies ahighly sinuousvalley, incisedby<3mintoaflat up to 500mwide alluvialfloodplain ofweakly cemented gravelscapped by sands and silts. Canyon and gorge reaches often lack thealluvial floodplain instead comprising bedrock reaches with some-times thin and transient alluvial covers. Flood hydrology is influencedby a semi-arid mountain climate with a marked altitudinal zonationwhere annual precipitation varies from 200 mm upstream (M'smerirvillage) to 150 mm downstream (Boumalne du Dad�es town) (Schulzet al., 2008; Dłu _zewski et al., 2013) (Fig. 3). Average daily dischargesare 33.3 m3/s but with seasonal variation linked to annual wintere-spring precipitation in drainage divide regions derived from Atlanticlow pressure incursions and rarer convective storm events from thetropics (e.g. Fink and Knippertz, 2003). The ephemeral tributariescommonly store and supply large volumes of coarse clastic sedimentto alluvial fans in theDad�es River valley dependent upon the tributarycatchment bedrock lithology, stratigraphy and structure. The fansbuild out onto the valley floor and are often eroded through interac-tionwith low frequency-highmagnitudeflooding (Stokes andMather,2015). Bedrockweathering and sediment supply are further enhancedby the absence of vegetation cover.

Throughout the studyarea, alluvial terraces are observed along thesides of the Dad�es valley and its larger tributaries. Their only docu-mentation is on recent geological maps (e.g. Carte G�eologique duMaroc, 1993) where up to 5 levels (q1eq5) are intermittently recor-dedusing the classical stratigraphic subdivisionnomenclature (Plaziatet al., 2008): Younger Quaternary ¼ Soltanien [q1]; MiddleQuaternary ¼ Tensiftien [q2]; Older Quaternary ¼ Amirien [q3],Saletien [q4], Moulouyen [q5]. The maps lack detail on mappingprocedure, height information, age assignment and any terraceenvironmental significance. During the Quaternary, arid climateconditions dominate, fluctuating between cool and dry glacials andwarm and dry interglacials (Lamb et al., 1994; Valero-Garc�es et al.,1998). Despite the aridity, periods of humidity with elevated precip-itation existed, becoming regionallyelevatedduring themajor climatetransitions (Tjallingii et al., 2008). During the glacials high relief areas(2e4 km) saw plateau icefield, valley glacier development, periglacialactivity (Wiche, 1953; Hughes et al., 2011) and fluctuating highmountain lake levels (Lamb et al., 1994; Valero-Garc�es et al., 1998);contrasting with enhanced aeolian dune activity in low relief andlower latitude western Sahara Desert regions (Lancaster et al., 2002).

3. Approach and methods

3.1. Terrace mapping, stratigraphy and sedimentology

River terraces were analysed using integrated field and remotesensing approaches in-between the towns of Boumalne du Dad�es

(downstream) and M'smerir (upstream) (Fig. 3). Remote sensingutilised 1 arc second SRTM derived digital elevation data, GoogleEarth imagery, topographic and geological maps collectivelyvisualised and interrogated in Arc GIS.

River terraces were identified using standardmorphological andgeological criteria (Stokes et al., 2012a; Mather et al., 2017). Terracesurfaces were often unclear due to burial by slope/fan deposits butvalley side erosion commonly revealed a flat to gently dippingdownstream terrace surface contact with overlying slope material.Terrace risers were normally steep and cliff-like, whilst terracebases revealed sharp lithological contrasts between the fluvialconglomerate and underlying bedrock. Contacts between con-glomerates and overlying slope deposits utilised differences insediment textures, fabrics and stratigraphy (Mather et al., 2017).

Terraces were mapped using a Trimble Geo-XH GPS to recordbreaks and changes in slope associated with inner valley terracemargins (i.e. closest to the modern river). Outer valley terrace mar-gins were normally estimated/extrapolated using slope morphologydifferences between the bedrock geology, the slope deposit and theterrace. Terrace heights were surveyed using a Trupulse 360B laserrangefinder, recording theelevationof the lowest pointof the terracebase above themodern river, in accordancewith other strath terracestudies in mountain belt settings (e.g. Stokes et al., 2012b). Terracebases were unclear in areas of Plio-Quaternary basin fill (e.g. FB andWTB settings) due to similarities in sediment textures, fabric andstructures. Here, terrace heights were based on survey estimatesusing differences in slope geomorphology (e.g. Mather et al., 2017).The locations and elevations of the terraces, together with those ofthe modern river long profile were compiled into an Excel database(Supplementary Information) and represented as a scatterplotterrace height-range diagram (Fig. 5). Using the database and theheight-range diagram, a terrace stratigraphic framework was con-structed. Normally, a terrace stratigraphic framework is constructedusing combinations of variables including: sedimentology, soils,terrace height, provenance, desert pavement etc. (e.g. Meikle et al.,2010; Stokes et al., 2012b). Here, the terrace stratigraphy was con-structed using terrace base elevation and assuming the simplestpalaeo-profile fit, since all other variables showed insufficientcontrast or are absent between levels. Terrace stratigraphic nomen-clature uses a letter and numbering system with T1 the lowest andyoungest level, with subsequent higher altitude/older terraces levelscorresponding to T2 and T3 etc. Terraces above T6 (50 m abovemodern channel) are rare fragmentary occurrences that make cor-relation difficult. These levels are referred to using height only butthe terrace numbering system allows future studies to assign furtherlevels (e.g. T7 etc.) if required.

Field mapping enabled type localities to be selected for detailedstudy, including sites with 1) well-preserved river terrace staircasesin contrasting geological-geomorphological contexts, 2) clear andrepresentative internal terrace stratigraphy and sedimentology, 3)suitable material for OSL sampling and 4) evidence for tectonicdeformation. Sites with clear and representative internal terracestratigraphy and sedimentology were described and logged usingsediment facies analysis (Miall, 1978) to enable sedimentary pro-cess and environment interpretations.

3.2. Terrace dating

Terrace chronology was established using OSL/IRSL dating ofquartz and feldspar, a technique routinely used for Quaternary riverterrace studies (e.g. Martins et al., 2010). Sands suitable for OSL/IRSLdating were extremely rare and were restricted to the T2 terrace(~10 m above the modern river) located only in the upstream studyarea within mid and proximal parts of the FTB region. Three siteswere sampled over a 10 km distance (Fig. 5; Supplementary

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 369

Information) to ensure stratigraphic continuity and to test thetemporal consistency of the T2 terrace age results. Sampling tar-geted fluvial sands that capped the coarse grained fluvial gravels(Fig. 9). These sands reflect overbank areas elevated adjacent to theactive channel (Section 5). After logging the sections, samplinglocations were prepared by removing weathered surface sedimentunder black out conditions. Steel tubes (~30 � 10 cm) werehammered into the sand units with tube ends sealed on removal, allunder blackout conditions. Host sediment water content wasvisually estimated in the field. A bulk sand sample was obtained forlaboratory dosimetry measurements.

The tube (180e250 mm sand fraction) and bulk samples wereprepared for laboratory dosimetry measurement using standardchemical cleaning preparation steps, followed by radionuclideconcentration measurement with final dose rate calculations usingthe conversions and calculations of Prescott and Hutton (1995) andAdamiec and Aïtken (1998) (Supplementary Information). A nom-inal water content of 5 ± 4% was used based on field estimates andlaboratory derived saturation moisture contents. Luminescencemeasurements were made using a Risø OSL reader (DA-20). Quartzluminescence followed the SAR protocol (Murray and Wintle,2000) involving natural quartz purity dose checks (Duller, 2003)and preliminary preheat plateau tests (SupplementaryInformation). Feldspar luminescence followed the infrared stimu-lated luminescence (IRSL) approach (Buylaert et al., 2012),measuring normal and post IRSL signals at low and high temper-atures (e.g. IR 50 and pIRIR 290). The SAR protocol (Thiel et al.,2011; Buylaert et al., 2012) was used for K-feldspar IRSL signalmeasurements and testing (Supplementary Information).

3.3. Terrace-rock strength relationships

The relationship between rock strength and terrace distribution

Mio-Quaternary

Oligo-MioceneCretaceous-Paleogene Mio-Quaternary Lower Juras

Miocenetaceous-PaleogeneOOOligOligoOligoOOligoOligogogo MM-Mioc-Mioc-MiocMiocenene

CCCreCreCretaCretaretaaccceoceou

Dist

Alti

tude

(m)

1500

1550

1600

1650

1700

1750

1800

1850

1900

1950

2000

2050

0 5 10 15 20 25

80m70m60m

140m

50m40m30m20m10m

Terr

ace

Aba

ndon

edM

eand

er

XXXX

XX

X

X

ThrustFront

Wedge-TopBasin

Fore

deep

Bas

in

<5mT1T2T3T4T5

n/a

Terr

ace

base

r evirevobat hgi ehT

yhpar git art secarr e

XXXX

T6

Gor

ge

Gor

ge

Dades Gorg

XX

StrongMixed-StrongMixed-Weak

Weak

Rock Strength

Stai

rcas

e

Fig. 5. River terrace, river gorge, bedrock and abandoned meander relat

was assessed using published geological maps and insitu fieldstrength testing. Within the study area, the Mesozoic and Cenozoicbedrock comprises a range of different sedimentary lithologies thatpossess different strength properties related to lithology textures(granular vs crystalline), cementation and discontinuity (joints,fractures, bedding, etc.) characteristics. Following Stokes andMather (2015) two strength-lithology types were identified basedupon (i) qualitative strength properties (above) and (ii) quantitativein situ mass strength measurements using a Schmidt hammer(Goudie, 2006):

Type 1 d massive crystalline limestone or cemented clasticsediments (conglomerates or sandstones) with Schmidt hammervalues of 40e60. Rock surfaces display minimal discolouration.Rock weathers/erodes into decimetre blocks, slabs or large clasts.

Type 2 d poorly cemented sandstone, siltstone, and mudstonewith Schmidt hammer values of <30. Rock surfaces are discoloured,and rock mass possesses a well-developed fissile fabric conduciveto granular weathering characterised by marked disintegration.

Stratigraphic units dominated by Type 1 or 2 lithologies wereclassified respectively as ‘Strong’ or ‘Weak’. Where a mixture ofType 1 and 2 lithologies occurred, then rock strength was classifiedas ‘Mixed-Strong’ or ‘Mixed-Weak’ depending on the Type 1 vs 2dominance. Table 1 summarises the stratigraphy, lithology and rockstrength relationships and downstream strength variability isplotted graphically alongside river terrace occurrences (Fig. 5).

4. River terrace distribution

Terraces were recorded at up to 140 m above the modern valleyfloor (Fig. 5). They are generally common features, often as laterallypersistent levels along valley sides at up to 50 m (T6) (Figs. 5 and6A,B), and then form infrequent fragmented levels and/or isolatedoccurrences at ~80 m and ~140 m (Figs. 5 and 6C). The lowest T1

sic Lower-Middle Jurassic Middle Jurassic

ance (km)30 35 40 45 50 55 60

Fold-Thrust Belt

e T. n

’ Dad

es G

orge

BedrockMeanders

Stai

rcas

e

OSL

OSL

OSL

ionships to the orogenic system, bedrock stratigraphy and strength.

Table 1Study area bedrock geology, stratigraphic configuration, distribution (Carte G�eologique du Maroc, 1975, 1990; 1993) and rock strength characteristics (Section 3). FTB ¼ Fold-Thrust Belt; WTB ¼ Wedge-Top Basin; TF ¼ Thrust Front, FB ¼ Foredeep Basin. Valley length distribution superscript a, b, c ¼ shared valley length of strike orientated valleyexploiting contact between different stratigraphic units.

Time Formation name Lithologies Orogenoccurrence

Study area valleylength distribution(km)

Rock strength

Cenozoic Miocene-Quaternary (M-Q) Aït Kandoula(AK)

Continental fluvial and fan conglomerate:cemented carbonate cobbles

WTB, FB 11.88 Mixed-Weak

Oligo-Miocene (O-M) Aït Ouglif (AO) Continental fluvial conglomerate,sandstone and siltstone

TF 2.1b Weak

Eocene (E) n/a Continental and marine limestone and marl TF 4.63a Mixed-StrongMesozoic Cretaceous (C) n/a Continental sandstone, shale and gypsum TF 6.89b Weak

Jurassic (J) Middle Bajocien Bin El Ouidane(BEO)

Marginal marine platform ooliticlimestone and marls

FTB 4.9 Mixed-Strong

Aalenien Azilal (A) Continental and marginal marine silts,marls, sandstone and dolomite

FTB 3.5 Weak

Lower Toarcien Tafraout (T) Continental fluvial and marginal marine/coastalplain marl, silt and sandstone.

FTB 3.2 Mixed-Strong

Pliensbachien Jbel Choucht(JC)

Massive marine platform limestone. FTB, TF 10.56a Strong

Pliensbachien Ouchbis (O) Interbedded deep marine rhythmiclimestone and marl.

FTB 17 Mixed-Weak

Fig. 6. River terrace field imagery: A) Laterally extensive T5 terrace levels (white arrows) buried by tributary fan slope sediments (black arrow), mid FTB region. B) T2-T3 terracelevel remnants, inset by large relict tributary fan, mid FTB region. C) The highest documented terrace level remnants at 140m, proximal FTB region. D) T1 terrace level (arrow) ~2 mabove modern river, immediately upstream of Tarhía n’ Dad�es gorge, distal FTB region. E) Well developed T1-T4 terrace staircase, downstream TF region. F) Abandoned bedrockmeander containing T2 upstream-downstream terrace remnants buried by locally derived slope deposits, proximal FTB region.

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379370

Fig. 7. Map (A) and topographic profile (B) of T1-T5 terrace staircase, upstream FTBregion terrace staircase. See Fig. 8E and Supplementary Information for OSL sampleinformation.

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 371

terrace level is slightly incised (<3 m) by the modern Dad�es Riverand occupies large areas of the valley floor, commonly used forsubsistence agriculture (Figs. 5 and 6D). Above T1 there is aconsistent vertical spacing of terraces up to 50 m at ~10 m intervals(T2 to T6) (Fig. 5). The stratigraphic relationship of the higherterrace levels above T6 (e.g. 50 m) is unclear and thus lacks cor-relation in this study. Google Earth and Excel databases of terracelocations, stratigraphy and bedrock geology-strength relationshipsare provided as Supplementary Information.

Levels T1 to T4 formwell preserved, extensive levels throughoutmuch of the FTB region (Fig. 5) where they overlie steeply dippingto flat lying Lower Jurassic limestone-mudstones of the OuchbisFormation in a wide, open valley (Figs. 3B and 6A,B). In proximal,upstream parts of the FTB there is a progressive increase in lime-stone as the bedrock geology changes to the Middle Jurassic (Figs. 2and 4A). Here, terraces form one of the best developed staircases inthe study area, comprising T1 to T5 that have developed on abedrock spur as part of a pronounced valley meander (Figs. 5 and7). Upstream of this staircase decametre thick limestone units ofthe Middle Jurassic Bin El Ouidane limestone dominate, with thevalley form changing to a high sinuosity knick zone reach ofbedrock meanders that lack terraces (Figs. 4A and 6C). Down-stream, in the most distal part of the FTB region, terraces are absentfrom the Dad�es and Tarhía n’Dad�es Gorges, where vertical fluvialincision and limited valley widening has occurred into strongLower Jurassic Jbel Coucht limestone (Figs. 4C and D and 5). Thesegorges form marked lithological knick zones on the modern riverprofile (Fig. 5 and Boulton et al., 2014). Within the FTB region arefour occurrences of km-scale abandoned bedrockmeanders (Fig. 5),associated with the T2 level located on the SE side of the rivervalley. T2 fluvial sediments occur at the upstream entry anddownstream exit points of the abandoned meanders (Fig. 6F). Themeander cutoffs are infilled with slope colluvium but tributarystream incision through this fill often reveals the terrace and itsbasal Jurassic bedrock contact.

Within the WTB region terraces are unclear due to the litho-logical similarity with the Plio-Quaternary basin infill. Slopemorphological changes suggest some fragmentary T2 terrace oc-currences and an isolated T4 (Fig. 5).

Terraces become more common throughout the TF (Fig. 5).Upstream areas show a well-developed staircase comprising T1through T4 developed onto steeply dipping Neogene bedrock.Downstream is a more open valley, with a less well developed T1-T3 terrace staircase developed onto steeply dipping Cretaceousbedrock (Fig. 6E). Mid and downstream TF areas comprise rivergorge reaches cut into Jurassic and Palaeogene limestone lacking inriver terraces (Fig. 4G).

Within the FB, terraces are again unclear due to the lithologicalsimilarities with the Plio-Quaternary basin infill. Slope morpho-logical changes suggest fragmentary occurrences of T3. These ter-races are inset into a top basin fill surface (Fig. 3B).

The bedrock geology strength (Table 1) shows a relationship tovalley morphology (Figs. 3B and 4) and terrace occurrence (Fig. 5).High strength bedrock (Table 1) lacks terraces, forming narrowvalleys with river gorge and canyon formation (Figs. 4A, C, D and6C). Weak and Mixed-Weak bedrock are associated with widervalley forms and terrace occurrence, with well-developed T2-T4levels along the lower valley sides but lacking higher-older levels.Mixed-Strong bedrock is often associated with the beginnings ofnarrower valley canyon development and the most well-developedterrace staircases (Figs. 5 and 7).

5. River terrace sedimentology and stratigraphy

River terraces are characterised by 1e3 m and rare 5e7 m thick

conglomerate units (Figs. 8 and 9). The conglomerates are coarsegrained and dominated by boulder-cobble size clasts of Mesozoiclimestone that are set within a poor-moderately sorted matrix oflimestone pebbles, gravel and sand (Fig. 8A and B). The largestclasts are well rounded and often display clear imbrication(Fig. 8A and B). The conglomerates are either massive or displayweak horizontal stratification, with rare occurrences of metrescale low angle cross-stratification (Fig. 8A,B, C,D). Terrace basalcontacts are sharp, revealing an angular unconformity with un-derlying tilted Mesozoic/Cenozoic sediments (Fig. 8A). The basesdisplay an undulating topography, whose scale is determined bythe dip and lithological variability of the underlying bedrock.Post-depositional carbonate cementation of clasts is common,especially at the terrace bases (Fig. 8A). Terraces are rarely cappedby fine to coarse sand units up to ~2 m thick that are massive,lacking sedimentary structures (Figs. 8E and 9). Sand grains aredominated by carbonate compositions but do contain somequartz and feldspar.

The terrace conglomerates are interpreted to have beendeposited by high energy fluvial processes based upon their coarsegrain size and down valley imbrication. Horizontal and low anglecross-stratification suggests deposition as gravel sheets and

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379372

longitudinal bar forms (lithofacies Gm, Gp: Miall, 1978) within abraided river system (e.g. Hein and Walker, 1977). The lack of finesand soil development within the conglomerates suggests highsediment mobility and reworking, as part of a sustained period ofvalley floor aggradation. Thickness variations in terrace bodies canbe accounted for by either local accommodation space variability orby valley morphological differences between the valley sides andpalaeo-axis/thalweg. Conglomerate carbonate cementation istypical of groundwater processes, with preferential cementationalong terrace bases due to permeability differences between con-glomerates and the bedrock (e.g. Nash and Smith, 2003). Theoverlying rare sands represent floodplain sedimentation or inoverbank areas when flood waters overtop channel margins withdeposition in slack water areas (e.g. Benito et al., 2003). Sedimen-tary structures are common in floodplain and slack water deposits(e.g. horizontal lamination or climbing ripples) but here theirabsence can be explained by insufficient grain size variabilitywithin the sediment laden flood waters. The quartz and feldsparcontent of the sands could be of aeolian Saharan dust origin but

Fig. 8. River terrace sedimentology. A) T6 terrace in the upstream TF region illustrating tcemented base (arrowed) and its sharp basal contact with underlying Neogene bedrock. B) TJurassic limestone. C) T2 terrace in mid FTB region showing thicker fluvial sediment accumsedimentology. E) Fine sands (FS) and slope colluvium (SC) capping the T2 terrace sedimesedimentology of slope deposits that bury river terraces throughout the study area, notingdownstream T3 abandoned meander loop, mid FTB region (Fig. 5).

more simply is reworked from Middle-Upper Jurassic continentalsedimentary bedrock found upstream of the study area.

Terrace conglomerates and rare sands are commonly overlain byup to 5m of coarse grained gravels (Figs. 8F and 9). These are poorlysorted and weakly stratified, with angular clasts whose composi-tion reflects the local bedrock geology. The gravels mantle thevalley sides in relict degraded patches and rarer continuous apronswith sediment building out over the river terraces, possessing aslope apron morphology. When observed in proximity to tributaryjunctions the gravels have a fan-surface like morphology. Sedi-ments are also associated with meander cut-offs, forming a slopeinfill of the abandoned valleys (Fig. 6F). These texturally immaturegravels are interpreted as localised gravity driven slope deposits. Attributary junctions, the gravels relate to debris flows as relicts oftributary junction fans, similar to those observed in the modernDad�es River (Stokes and Mather, 2015; Mather et al., 2017). Theterrace capping slope deposits are important for preserving theunderlying fluvial sediments, acting as a valuable stratigraphicmarker to assess the relative timings of river activity.

he typical rounded and imbricated cobble conglomerate sedimentology, a carbonate3 terrace in the upstream TF setting illustrated imbrication of rounded cobble clasts ofulation and m-scale low angle cross bedding (white box ¼ photo D). D) Detailed T2

nts (T2S), proximal FTB region. OSL sampling location arrowed (Section 6). F) Typicalpoor sorting, soil development in fines and gentle dip towards valley. Example from

Fig. 9. Logs of terrace sections used for OSL dating. See Fig. 5 and Supplementary Information for locations. Cgl ¼ cobble-boulder conglomerate/gravels; Gm ¼ fluvial gravels;Gms ¼ slope/tributary fan gravels; VFS ¼ very fine sand; CS-G ¼ coarse sand-gravel.

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 373

6. OSL/IRSL dating

Quartz and feldspar OSL/IRSL results are presented in Tables 2and 3 with methods and a detailed technical discussion providedas Supplementary Information. The quartz age estimates (Dad�es-1 ¼ 69 ± 7 ka; Dad�es 2 ¼ 78 ± 4 ka; Dad�es-3 ¼ 76 ± 5 ka) show amean age of 74 ka with a small 5 ka range (Table 2). Despite the ageconsistency between locations, older timescale quartz OSL ages(e.g. >60e70 ka) are often considered to underestimate absoluteages (e.g. Buylaert et al., 2007) and are thus treated as minimumvalues. The feldspar age estimates (Dad�es-1 ¼ 104 ± 14 ka; Dad�es2¼ 70± 6 ka; Dad�es-3¼102 ± 8 ka) reveal amean age of 92 kawitha range of 9 ka (Table 3). However, these estimates show somedegree of variability when compared to each other and their

Table 2Summary of equivalent dose values and calculated ages for quartz grains.

Sample Natural De ± se (Gy) Saturateda De ± se(Gy)

Dose Rate(Gy/ka)

Age (Ka)

Dades-1 190 ± 17 (n ¼ 17) 269 ± 29 (n ¼ 16) 2.76 ± 0.1 69 ± 7Dades-2 172 ± 5 (n ¼ 27) 327 ± 21 (n ¼ 19) 2.21 ± 0.1 78 ± 4Dades-3 195 ± 6. (n ¼ 28) 332 ± 18 (n ¼ 20) 2.55 ± 0.1 76 ± 5

a Signal saturation equivalent at 2 � Do (Wintle and Murray, 2006).

Table 3Summary of equivalent dose values and calculated ages for K-feldspar grains.

Sample De (Gy)IR50 ± se

De (Gy)pIRIR290* ± se

Ratio IR50: pIRIR290 Dose Rat

Dades-1 139 ± 24 (n ¼ 8) 371 ± 58 (n ¼ 5) 1: 1.4 3.6 ± 0.1Dades-2 114 ± 8 (n ¼ 9) 212 ± 10 (n ¼ 9) 1: 1.9 3.0 ± 0.1Dades-3 155 ± 24 (n ¼ 16) 343 ± 28 (n ¼ 16) 1: 1.7 3.4 ± 0.1

*Corrected for residual dose significant at 24Gy. No ages are presented for IR50. See sup

equivalent quartz OSL ages (see Supplementary Information fordiscussion).

7. Discussion

7.1. Geologic controls

Modelling studies suggest that bedrock benches beneath strathterraces formwhen there is a change in ratio from vertical incisionto lateral erosion (Hancock and Anderson, 2002). Within the studyarea, the potential for lateral bedrock erosion is greater in areas ofweaker bedrock and restricted in stronger bedrock (Fig. 10). This ismost notable in middle parts of the FTB (Fig. 5), where mixed-weakstrength interbedded limestone-mudstone of the Lower JurassicOuchbis Formation (Fig. 4B) displays well-developed and laterallycontinuous T1, T2, T3 and T4 levels over some 20 km (Fig. 5). Thestructural configuration of this bedrock into a syncline, and therouting of the Dad�es River down the syncline axis (Fig. 2) enhancefurther the potential for lateral erosion. Bedrock on both valleymargins dips towards the valley axis due to the fold limbarrangement (see ❷ on Fig. 3). This facilitates valley margin erosionthrough slope undercutting and shallow translational failures thatexploit bedding plane dip and discontinuity configuration (Matherand Stokes, in press), a process reported in bedrock erosion field

e ± se (Gy/ka) Corrected* Age (Ka) pIRIR290 Uncorrected Age (Ka) pIRIR290

104 ± 17 109 ± 1670 ± 6 77 ± 5102 ± 8 108 ± 7

plementary Information for discussion.

Fig. 10. A) River gorge reach of structurally thickened ‘strong’ limestone with noterrace formation. B) Confined canyon reach where river terrace staircase developmentis controlled by the strength variability in a dipping stratigraphic sequence of TafraouteFm mudstone (T-MSt) and limestone (T-Lst). C) Oblique Google Earth image illustratingrock strength, structure and stratigraphic sequence relationships to terrace staircaseformation at the Tilout site (Fig. 7).

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379374

and modelling studies (e.g. Weissel and Seidl, 1997).Fluvial bedrock erosion occurs through combinations of solu-

tion, plucking, cavitation and abrasion depending upon substratelithology and discontinuities (Whipple et al., 2000). Within thestudy area, the dominance of Mesozoic carbonate substrate sug-gests some degree of solution (e.g. de Jong et al., 2008). However,field evidence from bedrock reaches (Stokes et al., 2008) and thebasal contacts of terraces (Section 5) suggest erosion is dominatedby plucking/cavitation and abrasion. Terrace sediments are almostentirely composed of cobble size clasts of Lower-Middle Jurassiclimestone (Section 5) with little mudstone clast material. Similargrain size and composition characteristics occur within the activethin alluvial covers in the modern Dad�es River bedrock reaches.Limestone bedrock readily breaks down into decimetre blocks due

to joint configuration (Stokes et al., 2008). The massive crystallineproperties of the limestone provide high rock strength, acting as anabrasion tool when transported as bedload over weaker mudstonebedrock. In terms of strath terrace formation, the bedrock benchcorresponds to a period where bedload and discharge are sufficientto abrade and widen the valley floor but sediment supply isinsufficient to cause valley floor aggradation (Hancock andAnderson, 2002; Montgomery, 2004).

Terrace staircases record longer term patterns of fluvial down-cutting, with multiple ratio changes from vertical incision to lateralerosion. Studies routinely attribute terrace staircase formation tosustained base-level lowering driven by combinations of tectonicand/or climatic changes (e.g. Starkel, 2003; Bridgland andWestaway, 2008a,b). Within the study area, terrace staircasescomprising >4 levels occur in localised parts of the FTB and TFsettings (Fig. 5). These staircases occupy locations of pronounceddeviation in river valley direction, developing on valley marginspurs where the river has progressively migrated and incised(Figs. 7 and 10B, C). Whilst base-level lowering has undoubtedlyinfluenced the terrace staircase development, rock strength hasplayed an equally important role. Staircases have developed inareas of weaker rock strength (e.g. TF staircase ¼ Neogene mud-stones and sandstones) that are conducive to both vertical andlateral erosion. The FTB staircase (Figs. 5, 7 and 10B, C) is particu-larly notable for being the most complete in terms of the number oflevels (�5: T1-T5) and showing a clear relationship to substratestrength. Here, the Dad�es valley cuts through the Tafraout Forma-tion (Table 1), a Lower-Middle Jurassic sequence of limestone andmudstone (Carte G�eologique du Maroc, 1993). The terrace staircasehas locally exploited a mudstone part of the sequence along thelower and middle valley sides, whilst a return to decimetre thicklimestone in upper valley side positions has suppressed terraceformation at higher landscape levels (Fig. 10B and C). Downstream,inmiddle reaches of the FTB, terrace staircases may have developedbut have been eroded due to being subjected to longer timescales ofvalley side erosion into the underlying homogeneously weakbedrock (e.g. Ouchbis Fm). Thus, the collective strength of the un-derlying bedrock geology and its stratigraphic and structuralconfiguration can either enhance or suppress terrace formation orcontribute significantly to poor terrace preservation (Fig. 10),similarly observed for river gorge development in the area (Stokeset al., 2008).

7.2. Climate controls

In this study, the typical terrace stratigraphic configuration,irrespective of landform level, is a bedrock bench, overlain bycoarse fluvial conglomerates with rare caps of overbank sands thatare collectively buried by slope/tributary fan sediments (Fig. 11).This sequence can be explained by climate-related behaviours ofthe river channel and its adjacent hillslopes (Fig. 11). The fluvialconglomerates correspond to the aggradation phase (Fig. 11), whenthere is elevated coarse-grained sediment supply to the valley floorfrom the hillslopes, tributaries and upstream catchment areas dueto effective slope-channel coupling (sensu Harvey, 2002) and suf-ficient discharge to mobilise coarse sediment (e.g. Bull, 1991). Theoverlying overbank sands can be explained in two ways: 1) ascontinued valley floor aggradation, but the final stages immediatelyprior to the onset of incision, where coarse-grained sedimentsupply is limited, or 2) in relation to valley floor incision (Fig. 11)when the river channels have a reduced sediment supply, butwhere floodwaters can still overtop incised terrace surface (e.g.Wang et al., 2014). The latter is the simpler (and preferred) expla-nation being commonly observed in the modern Dad�es River floodhydrology where flood waters overtop the valley floor T1 terrace

Fig. 11. A conceptual river terrace-climate model for the Dad�es River illustrating therelationship between glacial-interglacial climates, valley floor incision-aggradationpatterns and river channel-hillside coupling status.

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 375

depositing sand and silt. The terrace capping slope/fan depositssuggest valley floor incision (Fig.11) with hillslopematerial buryingthe valley side terrace remnants but with insufficient river flooddischarge to activate meaningful valley floor aggradation, adecoupling of the channel and hillslopes (Mather et al., 2017). Thebedrock bench is cut early in the transition between the incisionand aggradation phases (Fig. 11), a threshold period when floodregime and sediment supply is increasing to a sufficient level forbedload transport (tools) to cause lateral erosion into valley sides(Hancock and Anderson, 2002; Montgomery, 2004).

OSL/IRSL dating of the T2 overbank sands, although limited innumber, provides some insight into terrace incision timing (Fig. 11).Whilst quartz OSL ages were considered as minimum ages only(Table 2; Supplementary Information), the corrected feldspar ages(Table 3) of 104 ± 17 ka (upstream), 70 ± 6 ka (downstream) and102 ± 8 ka (mid-reach) were consideredmore representative. Theseages, including error bars, span the Marine Isotope Stage (MIS) 5interglacial andMIS 4 glacial climate periods. River terrace research(e.g. Vandenberghe, 2015 and references therein) states that inci-sion normally takes place during the climate transitions. This studyagrees, with T2 incision most likely to have taken place during theMIS 6/5 cold-warm transition. By stratigraphic bracketing, thismeans that aggradation of the T2 fluvial gravels probably took placeduring the MIS 6 glacial and that slope colluvium was depositedover the T2 gravels and rare overbank sands during the MIS 4-2glacial. The T2 stratigraphic configuration and timing has someaffinities to sequences described by Weisrock et al. (2006) andMercier et al. (2009) from the coastal region of the Anti-AtlasMountains (Fig. 1). Using OSL, U-Series and radiocarbon dating,they proposed a major pre-MIS 5 fluvial incision and basal fluvialgravel aggradation phase, followed by MIS 5-3 fluvial-slope trav-ertine-silt-sand aggradation and capped by a MIS 2 aeolian-slopesilt deposit. Furthermore, studies of western Mediterranean datedterrace records (e.g. Macklin et al., 2002; Santisteban and Schulte,2007) suggest a complex pattern of fluvial aggradation and inci-sion phases during the Middle-Late Pleistocene. Of note are reportsof regionally synchronous MIS 6 aggradation and MIS 5 incisionphases in southern Spain and NE Libya (Macklin et al., 2002). Thus,it seems that on the south side the High Atlas along the NWSaharanmargin, the MIS 6 fluvial aggradation and onset of incision at the

MIS 6/5 transition have some degree of harmony with the WesternMediterranean region, acknowledging the restricted dataset interms of sites and age control. Dated fluvial sequences are startingto emerge from northern Morocco in relation to early humanoccupation research, describing fluvial aggradation inMIS 6/5e (e.g.Bartz et al., 2015). However, these are localised site specificarchaeologically focussed studies where the fluvial geomorpho-logical context is insufficient to make meaningful comparisons toriver terrace records.

If a similar climate relationship is considered for T1 thatcurrently occupies large parts of the wider valley floor settings, thismeans that the coarse grained fluvial sediment corresponds to theLate Pleistocene glacial (MIS 2), whilst the onset and ongoing(small) incision phase is Late-Pleistocene-Holocene (MIS 2/1).Sands cap the T1 terrace fluvial gravels and these are similar to therare occurrences of the OSL/IRSL dated T2 sands. Interestingly, slopecolluvium is not yet building out over the T1 terrace, with terracetop sedimentation limited to localised tributary junction fan oc-currences (Stokes and Mather, 2015). Much of the valley sides arebedrock, with only patches of colluvial veneer remaining. Thissuggests hillslope sediment is almost exhausted and decoupledfrom the river channel (Fig. 11) with another phase of cold-climate(peri)glacial weathering required to regenerate sufficient hillslopematerial to start to bury the T1 terrace (Mather et al., 2017).Although glacial periods are dominated by aridity in NW Africa asillustrated by Saharan dune activity (Fig. 1; e.g. Lancaster et al.,2002), high altitude lake records (e.g. Lake Isli [Fig. 1.], ~50 kmnorth of the Dad�es: Lamb et al., 1994; Valero-Garc�es et al., 1998)show complex and detailed patterns of cold climate aridity-humidity variability which helps clarify the timing of the T1terrace formation. The lake record shows a marked change fromelevated aridity, low lake levels (>35e28 ka) and erosion(28e20 ka) during the Full Glacial, changing to sustained humidityand high lake levels (20e11 ka) during the late Glacial Transition(Valero-Garc�es et al., 1998). Lake Isli is positioned on the Dad�esRiver drainage divide and thus the downstream valley effects areprobably slope colluvium activity supplying excess sediment to thevalley floor resulting in aggradation during the cold-arid FullGlacial (pre-20 ka), followed by slope erosion and valley floorincision during the late Glacial Transition period (post-20 ka) whenthe climate ameliorates becoming more humid (e.g. Gibbard andLewin, 2002). Furthermore, although the Dad�es catchment wasnot glaciated, the development of small ice fields and valley glacieradvance in the Toubkal region (Fig. 1) at ~76, 24 and 12 ka (Hugheset al., 2011) further illustrates the cold climate conditioning of theHigh Atlas landscape. In the Dad�es catchment, periglacial activitywould have been important for slope weathering and sedimentsupply to the valley floor but clear periglacial evidence are lackingfrom the terrace sediments (e.g. cryoturbations etc). The 24 kaToubkal advance approximates to the proposed T1 aggradationperiod, but the relationship of the 76 ka and 12 ka events to theterrace record remains unclear.

If the main periods of valley floor fluvial aggradation took placein the glacial periods (i.e. MIS 2 and 6) and major incision duringwarming transitions (i.e. 6/5 and 2/1), this suggests that fluviallandscape dynamics in the south-central High Atlas are tuned to100 ka Milankovitch cycles, showing affinity to other terrace-climate studies from around the world (e.g. Bridgland andWestaway, 2008a; Vandenberghe, 2015). If this Milankovitch rela-tionship is developed further, and that each of the well-developed,laterally continuous terrace level (T1-T4) relates to a full climatecycle, it means that the Dad�es River records some 400 ka of fluvial-landscape dynamics spanning the Middle-Late Pleistocene (Fig. 12).It also suggests that rare terraces higher than T6, i.e. >50 m abovethe modern river, are Early Pleistocene in nature (Fig. 12) and they

T1T2

T3T4

T5T6

0102030405060708090

140130130120110100

T7?T8?

T9?

T15?

1 ma

500 ka

100 ka0

1.4 ma

Late

Mid

dle

Early

Hei

ght a

bove

mod

ern

river

bed

(m)

PLEI

STO

CEN

E

Fig. 12. Pleistocene temporal pattern of the Dad�es River terrace sequence.

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379376

offer a significant potential for linking orogenic basin sedimentaryrecords (e.g. WTB, FB) to incision of the FTB region.

High Atlas drainages, including the Dad�es River, are routeddownstream into the Ouarzazate Basin. Here, remnants of largealluvial fans are recorded in west-central basin areas, with studiesdocumenting up to 5 inset fan surfaces (St€ablein, 1988; Littmannand Schmidt, 1989; S�ebrier et al., 2006). This number of surfacelevels has some broad stratigraphic agreement with the upstreamDad�es River terraces (e.g. T1eT5). Furthermore, there is some de-gree of temporal agreement from cosmogenic exposure dating ofthe fan surfaces by Arboleya et al. (2008). They suggest that fansurface abandonment and incision is climatically controlled,occurring broadly over a series of 100 ka cycles during the Holocene(MIS 1) and Middle-Late Pleistocene (MIS 5, 7, 9). Delcaillau et al.(2016) have described fan-terrace sequences from the Ourika val-ley on the north (Marrakech) side of the High Atlas. They considerclimate as a key control on the incision and aggradation patternsbut their study lacks age control, and thus clear climate relation-ships are lacking.

This broad pattern of 100 ka cycles is illustrated in regional scaleQuaternary climatemodelling studies for NWAfrica, where markedspatial and temporal patterns of changes in precipitation arerecorded (Tjallingii et al., 2008). For the Sahara, aridity dominatesbut precipitation becomes elevated during the climate transitions:1) interglacial-glacial (e.g. MIS 5-4) and 2) glacial-interglacial (MIS2-1) (Tjallingii et al., 2008), coinciding with the proposed periods ofincision in the Dad�es River.

7.3. Base-level lowering controls

Whilst climate ultimately controls variations in sediment-watersupply leading to changes in valley floor aggradation and incision(Bull, 1991), sustained base-level lowering is required to form aninset sequence of river terrace levels (Vandenberghe and Maddy,2001; Bridgland and Westaway, 2008a). If the rate of base-levellowering is small then complex fill terraces develop (Lewin andGibbard, 2010), contrasting with high rates of base-level fallwhich leads to strath terrace and staircase formation (Starkel,2003). Base-level lowering can be controlled by combinations ofclimate, river capture and tectonics.

Climate-related base-level lowering is normally attributed toeustacy, typically affecting downstream, coastal regions of riversystems where falling sea level causes incision, whereas rising sealevel causes aggradation (e.g. Stokes et al., 2012a and referencestherein). The continental situation of the Dad�es River (~900 km

from modern sea level) means that eustatic base-level loweringaffects are negligible.

River capture involves drainage rerouting and catchment areaexpansion and loss respectively of the capturing and capturedsystem (e.g. Stokes et al., 2002). The enlarged capturing system is ata lower base-level and once capture occurs, the base-level differ-ence is transmitted upstream through the captured catchment(Mather et al., 2002). The Ouarzazate Basin has been captured bythe Draa River in the Quaternary, whereby the internally drainedOuarzazate Basin was captured and re-routed southwards acrossthe Anti-Atlas through the Draa Gorge (St€ablein, 1988; Arboleyaet al., 2008). Capture-related impacts upon the drainage networkare unclear due data collection absence. However, capture-relatedincision should account for some fluvial erosion in lower drainagesystem parts of the Ouarzazate Basin. Whether capture-relatedincision has transmitted upstream across the SAF and upstreaminto the High Atlas FTB region is unclear but unlikely (Boulton et al.,2014).

Tectonic uplift is a commonly cited mechanism for long termbase-level lowering in river terrace studies (e.g. Lav�e and Avouac,2001; Cunha et al., 2008), with fill terraces forming under lowuplift rate conditions and strath terraces forming under interme-diate and higher uplift rate scenarios (Starkel, 2003).

The High Atlas topography has been generated during theCenozoic linked to ongoing Africa-Europe collision. Despite debate,relief appears to have been generated during the Eocene-Oligoceneand Plio-Quaternary (Frizon de Lamotte et al., 2000), throughcombined mechanical shortening (folding/thrusting) and mantle-related isostatic processes (e.g. Teixell et al., 2003; Missenardet al., 2006). The Dad�es terrace record corresponds to the Plio-Quaternary uplift stage, recording the most recent phase oflonger term drainage evolution. The uplift has generated a regionalNNW-SSE topographic gradient but the Dad�es River is routedobliquely SW across this gradient exploiting the axes of km-scalefold structures within the Jurassic bedrock of the FTB region(Figs. 2 and 4). Fig. 5 reveals a marked parallel configuration of theT1 to T4 terrace levels as they pass through all tectonic componentsof the High Atlas orogen. This includes the TF regionwhere elevatedrates of active tectonic uplift could be expected. Field mapping andcross-section compilation by Tes�on and Teixell (2008) demon-strates extensive folding and faulting in the TF region but this isrestricted to Jurassic to Plio-Quaternary bedrock. Our terracemapping did reveal minor m-scale thrust faulting of terrace sedi-ments but this was exceptionally rare, affecting the T3 level(Fig. 13). Faulting exploited bedding planes of tilted Neogenebedrock, with propagation up into the terrace sediments towardsthe NW (Fig. 13). Fault planes were low angle (~30�), with small(<2 m) vertical and horizontal offsets (Fig. 13). This lack of terracedeformation, even in TF areas where it would be expected, suggeststhat the High Atlas in the Dad�es River region was characterised bylow rate tectonic activity, at least from the Middle-Late Pleistoceneonwards. This is further supported by low modern seismic activity,where earthquakes, concentrated along the SAF region, are infre-quent and low magnitude (magnitude <4.9) (Medina andCherkaoui, 1991). However, there is uncertainty as to how defor-mation at the surface relates to the deformation at depth whichgenerates earthquakes (S�ebrier et al., 2006). Quaternary fan sur-faces south of the SAF in the Ouarzazate Basin do display a greaterdegree of surface faulting and folding (S�ebrier et al., 2006; Arboleyaet al., 2008) but these are spatially variable along different faultsegments of the SAF mountain front. However, this surface defor-mation requires improved age control to confirm deformationtiming and its relationship to basin and mountain front incision.

River terraces with age control can be used to quantify fluvialincision rates which in turn can be developed further to inform on

Fig. 13. T3 terrace deformation in the Thrust Front region (31�26058.92N5�58012.63W), illustrating thrust faulting (dip ¼ 30�; apparent strike ¼ 065) of fluvialterrace gravels. Faulting is exploiting bedding plane dip of tilted Neogene bedrock (45/065 NW).

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 377

spatial and temporal patterns of uplift (e.g. Lav�e and Avouc, 2001).In this study, the OSL dated fluvial overbank sands of the T2 terraceprovide some opportunity for quantifying incision. The Dad�es-1(104 ± 17 ka) and Dad�es-3 (102 ± 8ka) samples were taken fromsands immediately overlying coarse fluvial gravels at heights of12.5 m and 17.75 m above the modern river level (Fig. 10), pro-ducing respective incision rate estimates and error ranges of 0.12(0.10e0.14) mm a�1 to 0.17 (0.16e0.19) mm a�1. Higher rates of0.3e1.0 mm a�1 were reported by Arboleya et al. (2008) fromincised Quaternary fans in the Ouarzazate Basin, possibly reflecting1) proximity to the Dad�es-Draa capture site, 2) positioning along amore active SAF thrust segment or 3) rock strength differencesbetween different components of the orogen.

Compared to other studies that report incision rate from riverterraces in a collisional mountain tectonic context (e.g. Lav�e andAvouac, 2001; Cunha et al., 2008) the rates reported here are low,suggesting the south-central High Atlas has evolved under low rockuplift rate conditions since the Late Pleistocene. Furthermore, thesystematic altitudinal spacing of the T1-T4 terraces throughout the60-km-long study area suggests that similarly low incision rates(<0.2 mm a�1) have been maintained throughout the Quaternary,at least from the Middle Pleistocene onwards. The consistency inspatial and temporal distribution of the river terraces implies thatthe Dad�es River is responding to some kind of regional tectonicallydriven, low rate base-level lowering process. Rather than fault/foldrelated mechanical uplift, the base-level lowering could be linkedto isostatic uplift related to mantle upwelling under the High Atlasthat has been widely associated with longer term geophysical-seismic studies of High Atlas topographic development (e.g.Missenard et al., 2006) and from regional quantitative tectonicgeomorphological studies of drainage networks along the northand south of the central High Atlas (Delcaillau et al., 2010; Boultonet al., 2014).

8. Conclusions

Collectively, this study highlights the significant opportunitythat river terrace sequences hold for providing insights into Qua-ternary fluvial landscape development in NW Africa along theSaharan Desert margins. The study shows:

1. river terraces are widespread, yet little studied landformsthroughout the High Atlas and the broader NW African desertmountain landscapes, a notable research gap of significant

relevance for the geological, geomorphological and Quaternaryscience communities;

2. terrace formation in the High Atlas, and more generally withincollisional mountain belts, can be strongly influenced by rockstrength and its relationship to the structural and stratigraphicconfiguration of the orogen;

3. dating of river terraces within dryland mountain landscapes inNW Africa is possible despite the rareness of fluvial sands withappropriate mineral compositions suitable for dating longertime series (e.g. OSL);

4. with some age control, terrace formation can be linked toclimate-related incision and aggradational behaviour of theriver channel and valley side slopes, operating over 100 kaclimate cycles. For a given cycle, incision occurs during the cold-warm transitions and aggradation during the full glacial. This isa pattern recorded elsewhere along the NW Saharan margin inthe area south of the High Atlas, and appears synchronous withwestern/southern Mediterranean fluvial archives;

5. the terrace sequence suggests that fluvial landscape develop-ment spans the Late-Middle Pleistocene and possibly into theEarly Pleistocene. This provides a significant future opportunityfor a) understanding the Cenozoic development of the HighAtlas Mountain topography through linkage and integration ofgeological and geomorphological records; and b) understandingspatial and temporal patterns of low latitude Quaternary climatechange through integration with other limited regional climatearchives (e.g. glaciers, lakes, dunes etc.) from the NW Saharanmargin;

6. the spatial and temporal uniformity of terrace occurrence sug-gests a very low regional incision rate across all components ofthe High Atlas orogen (i.e. FTB, WTB, TF and FB);

7. the low regional incision rate coupled with a general absence ofterrace tectonic deformation suggests that base level lowering isprobably driven by mantle-related rock uplift rather than me-chanical fold and thrust related crustal shortening. This furtherimplies that High Atlas relief generation is an Early Pleistoceneor older phenomenon.

Acknowledgements

Funding: This work was supported by a National Geographicresearch grant (8609-09). Jamie Quinn is thanked for drafting Fig. 2.We thank Profs. David Bridgland and Jef Vandenberghe for theirpositive and thought provoking reviews and QSR Editor Prof Yangfor editorial handling.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.quascirev.2017.04.017.

References

Adamiec, G., Aïtken, M.J., 1998. Dose-rate conversion factors: update. Anc. TL 16,37e50.

Arboleya, M.-L., Babault, J., Owen, L.A., Teixell, A., Finkel, R.C., 2008. Timing andnature of Quaternary fluvial incision in the Ouarzazate foreland basin, Morocco.J. Geol. Soc. Lond. 165, 1059e1073.

Babault, J., Van Den Driessche, J., Teixell, A., 2012. Longitudinal to transversedrainage network evolution in the High Atlas (Morocco): the role of tectonics.Tectonics 31, TC4020.

Bartz, M., Klasen, N., Zander, A., Brill, D., Rixhon, G., Seeliger, M., Eiwanger, J.,Weniger, G.-C., Mikdad, A., Brückner, H., 2015. Luminescence dating ofephemeral stream deposits around the Palaeolithic site of Ifri n'Ammar(Morocco). Quat. Geochronol. 30, 460e465.

Benito, G., S�anchez-Moya, Y., Sope~na, A., 2003. Sedimentology of high-stage flooddeposits of the Tagus River, Central Spain. Sediment. Geol. 157, 107e132.

Boulton, S.J., Stokes, M., Mather, A.E., 2014. Transient fluvial incision as an indicator

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379378

of active faulting and Plio-Quaternary uplift of the Moroccan High Atlas. Tec-tonophysics 633, 16e33.

Bridgland, D.R., Westaway, R., 2008a. Climatically controlled river terrace staircases:a worldwide Quaternary phenomenon. Geomorphology 98, 285e315.

Bridgland, D.R., Westaway, R., 2008b. Preservation patterns of Late Cenozoic fluvialdeposits and their implications: results from IGCP 449. Quat. Int. 189, 5e38.

Bridgland, D.R., Westaway, R., 2014. Quaternary fluvial archives and landscapeevolution: a global synthesis. Proc. Geol. Assoc. 125, 600e629.

Bull, W.B., 1991. Geomorphic Responses to Climatic Change. Oxford Univ. Press,Oxford, p. 326.

Buylaert, J.P., Vandenberghe, D., Murray, A.S., Huot, S., De Corte, F., Van denHaute, P., 2007. Luminescence dating of old (~70 ka) Chinese loess: a compar-ison of single-aliquot OSL and IRSL techniques. Quat. Geochronol. 2, 9e14.

Buylaert, J.P., Jain, M., Murray, A.S., Thomsen, K.J., Thiel, C., Sohbati, R., 2012.A robust feldspar luminescence dating method for Middle and Late Pleistocenesediments. Boreas 41, 435e451.

Carte G�eologique du Maroc, 1975. Jbel Saghro-Dad�es. 1:200,000.Carte G�eologique du Maroc, 1985. Morocco map 1:1000,000. Notes and Memoir, No.

260.Carte G�eologique du Maroc, 1990. Zawyat Ahançal. 1:100,000.Carte G�eologique du Maroc, 1993. Tinerhir. 1:100,000.Cavini, L., 2012. Alluvial Deposits in Aït Seddrat Basin and Dad�es River Valley in the

Framework of the Evolution of the Southern Front of the Central High Atlas,Morocco. MSc Thesis. University of Florence, p. 27.

Chellai, E.H., Perriaux, J., 1996. Evolution geodynamique d'un bassin d'avant-pays dudomaine atlasique (Maroc): exemple des depots neogenes et quaternaires duversant septentrional de l'Atlas de Marrakech. Comptes Rendus 322, 727e734(Serie II).

Cunha, P.P., Martins, A.A., Huot, S., Murray, A., Raposo, L., 2008. Dating the Tejo riverlower terraces in the R�od~ao area (Portugal) to assess the role of tectonics anduplift. Geomorphology 64, 271e298.

Delcaillau, B., Laville, E., Amhrar, M., Namous, M., Dugu�e, O., Pedoja, K., 2010.Quaternary evolution of the marrakech high atlas and morphotectonic evidenceof activity along the Tizi N'Test fault, Morocco. Geomorphology 118, 262e279.

Delcaillau, B., Dugu�e, O., Namous, M., Pedoja, K., Amhrar, M., Laville, E., 2016.Pleistocene fluviatile deposits in the Ourika drainage basin (Marrakech HighAtlas, Morocco): indicators of climatic variations associated with base levelchange. Z. Geomorphol. 60, 1e20.

DeCelles, P.G., Giles, K.A., 1996. Foreland basin systems. Basin Res. 8, 105e123.de Jong, C., Cappy, S., Finckh, M., Funk, D., 2008. A transdisciplinary analysis ofwater

problems in the mountainous karst areas of Morocco. Eng. Geol. 99, 228e238.Dewey, J.F., Helman, M.I., Turco, E., Hutton, D.H.W., Knott, S.D., 1989. Kinematics of

the western mediterranean. In: Coward, M.P., Dietrech, D., Park, R.G. (Eds.),Alpine Tectonics, vol. 45. Geological Society, London Special Publication,pp. 265e283.

Dłu _zewski, M., Krzemie�n, K., Rojan, E., Biejat, K., 2013. Stream channel developmentin the southern parts of the High Atlas Mountains, Morocco. Geografija 49,10e21.

Duller, G.A.T., 2003. Distinguishing quartz and feldspar in single grain luminescencemeasurements. Radiat. Meas. 37, 161e165.

El Harfi, A., Lang, J., Salomon, J., Chellai, E.H., 2001. Cenozoic sedimentary dynamicsof the Ouarzazate foreland basin (central high atlas mountains, Morocco). Int. J.Earth Sci. 90, 393e411.

Fink, A.H., Knippertz, P., 2003. An extreme precipitation event in Southern Moroccoin Spring 2002 and some hydrological implications. Weather 58, 377e387.

Frizon de Lamotte, D., Saint Bezar, B., Brac�ene, R., Mercier, E., 2000. The two mainsteps of the Atlas building and geodynamics of the western Mediterranean.Tectonics 19, 740e761.

Frizon de Lamotte, D., Zizi, M., Missenard, Y., Hafid, M., El Azzouzi, M., Maury, R.C.,Charri�ere, A., Taki, Z., Benammi, M., Michard, A., 2008. The atlas system. In:Michard, A., Saddiqi, O., Chalouan, A., Frizon de Lamotte, D. (Eds.), ContinentalEvolution: the Geology of Morocco. Structure, Stratigraphy and Tectonics of theAfrica-Atlantic-mediterranean Triple Junction, vol. 116. Lecture Notes in EarthSciences, pp. 133e202.

Gibbard, P.L., Lewin, J., 2002. Climate and related controls on interglacial fluvialsedimentation in lowland Britain. Sediment. Geol. 151, 187e210.

Gomez, F., Beauchamp, W., Barazangi, M., 2000. Role of the atlas mountains(northwest Africa) within the African-Eurasian plate-boundary zone. Geology28, 775e778.

G€orler, K., Helmdach, F.F., Gaemers, P., Heissig, K., Hinsch, W., M€adler, K.,Shwarzhans, W., Zucht, M., 1988. The uplift of the central High Atlas as deducedfrom Neogene continental sediments of the Ouarzazate province, Morocco.Lect. Notes Earth Sci. 15, 363e404.

Goudie, A.S., 2006. The Schmidt Hammer in geomorphological research. Prog. Phys.Geogr. 30, 703e718.

Hancock, G.S., Anderson, R.S., 2002. Numerical modeling of fluvial strath-terraceformation in response to oscillating climate. Geol. Soc. Am. Bull. 114, 1131e1142.

Harvey, A.M., 2002. Effective timescales of coupling within fluvial systems. Geo-morphology 44, 175e201.

Hein, F.J., Walker, R.G., 1977. Bar evolution and development of stratification in thegravelly, braided, kicking Horse river, British Columbia. Can. J. Earth Sci. 14,562e570.

Hughes, P.D., Fenton, C.R., Gibbard, P.L., 2011. Quaternary glaciation in the atlasmountains of North Africa. In: Ehlers, J., Gibbard, P.L., Hughes, P.D. (Eds.), De-velopments in Quaternary Science, vol. 15. Elsevier, Amsterdam, The

Netherlands, pp. 1065e1074.Lamb, H.F., Duigan, C.A., Gee, J.H.R., Kelts, K., Lister, G., Maxted, R.W., Merzouk, A.,

Niessen, F., Tahri, M., Whittington, R.J., Zeroual, E., 1994. Lacustrine sedimen-tation in a high altitude, semi-arid environment: the palaeolimnological recordof Lake Isli, High Atlas, Morocco. In: Millington, A.C., Pye, K. (Eds.), Effects ofEnvironmental Change in Drylands. Wiley, pp. 147e161.

Lancaster, N., Kocurek, G., Singhvi, A., Pandey, V., Deynoux, M., Ghienne, J.-F., Lo, K.,2002. Late Pleistocene and Holocene dune activity and wind regimes in thewestern Sahara Desert of Mauritania. Geology 30, 991e994.

Lav�e, J., Avouc, J.P., 2001. Fluvial incision and tectonic uplift across the Himalayas ofcentral Nepal. J. Geophys. Res. 106, 26561e26591.

Leopold, L.B., Wolman, M.G., Miller, J.P., 1964. Fluvial Processes in Geomorphology.Freeman, San Francisco, p. 522.

Lewin, J., Gibbard, P.L., 2010. Quaternary river terraces in England: forms, sedimentsand processes. Geomorphology 120, 293e311.

Littmann, T., Schmidt, K.-H., 1989. The response of different relief units to climaticchange in an arid environment (southern Morocco). Catena 16, 343e355.

Macklin, M.G., Fuller, I.C., Lewin, J., Maas, G.S., Passmore, D.G., Rose, J.,Woodward, J.C., Black, S., Hamlin, R.H.B., Rowan, J.S., 2002. Correlation of fluvialsequences in the Mediterranean basin over the last 200 ka and their relation-ship to climate change. Quat. Sci. Rev. 21, 1633e1641.

Martins, A.A., Cunha, P.P., Buylaert, J.-P., Huot, S., Murray, A.S., Dinis, P., Stokes, M.,2010. K-feldspar IRSL dating of a Pleistocene river terrace staircase sequence ofthe Lower Tejo River (Portugal, western Iberia). Quat. Geochronol. 5, 176e180.

Mather, A.E., Stokes, M., 2017. Bedrock structural control on catchment-scale con-nectivity and alluvial fan processes, High Atlas Mountains, Morocco. In:Ventra, D., Clarke, L.E. (Eds.), Geology and Geomorphology of Alluvial andFluvial Fans: Terrestrial and Planetary Perspectives. Geological Society, London,Special Publications, p. 440 (in press).

Mather, A.E., Stokes, M., Griffiths, J.G., 2002. Quaternary landscape evolution: aframework for understanding contemporary erosion, Southeast Spain. LandDegrad. Dev. 13, 89e109.

Mather, A.E., Stokes, M., Whitfield, E., 2017. River terraces and alluvial fans: the casefor an integrated Quaternary fluvial archive. Quat. Sci. Rev. 166, 74e90.

Medina, F., Cherkaoui, T.-E., 1991. Focal mechanisms of the Atlas earthquakes, andtectonic implications. Geol. Rundsch 80, 639e649.

Meikle, C.D., Stokes, M., Maddy, D., 2010. Field mapping and GIS visualisation ofQuaternary river terrace landforms: an example from the Rio Almanzora, SESpain. J. Maps 2010, 531e542.

Mercier, N., Hatt�e, C., Fontugne, M., Reyss, J.-L., Valladas, H., Wengler, L., Brugal, J.-P.,Ouammoud, A., Weisrock, A., 2009. Chronology of Upper Pleistocene sequencesat Sidi Messaoud (wadi Noun, southwestern Morocco) based on 14C, opticaland U-series dating. Quat. Geochronol. 4, 326e334.

Miall, A.D., 1978. Lithofacies types and vertical profile models in braided rivers: asummary. In: Miall, A.D. (Ed.), Fluvial Sedimentology, vol. 5. Canadian Society ofPetroleum Geologists Memoir, pp. 597e604.

Michard, A., 1976. �El�ements de g�eologie marocaine. Notes M�em. Serv. G�eol. Maroc.252, 408.

Missenard, Y., Zeyen, H., Frizon de Lamotte, D., Leturmy, P., Petit, C., S�ebrier, M.,Saddiqi, O., 2006. Crustal versus asthenospheric origin of relief of the AtlasMountains of Morocco. J. Geophys. Res. 111, B03401.

Montgomery, D.R., 2004. Observations on the role of lithology in strath terraceformation and bedrock channel width. Am. J. Sci. 304, 454e476.

Murray, A.S., Wintle, A.G., 2000. Luminescence dating of quartz using an improvedsingle-aliquot regenerative-dose protocol. Radiat. Meas. 32, 57e73.

Nash, D.J., Smith, R.F., 2003. Properties and development of channel calcretes in amountain catchment, Tabernas Basin, southeast Spain. Geomorphology 50,227e250.

Plaziat, J.-C., Aberkan, M., Ahmamou, M., Choukri, A., 2008. The quaternary depositsof Morocco. In: Michard, A., Saddiqi, O., Chalouan, A., Frizon de Lamotte, D.(Eds.), Continental Evolution: the Geology of Morocco. Structure, Stratigraphyand Tectonics of the Africa-Atlantic-mediterranean Triple Junction, vol. 116.Lecture Notes in Earth Sciences, pp. 359e376.

Prescott, J.R., Hutton, J.T., 1995. Environmental dose rates and radioactive disequi-librium from some Australian luminescence dating sites. Quat. Sci. Rev. 14,439e448.

Santisteban, J.I., Schulte, L., 2007. Fluvial networks of the Iberian Peninsula: achronological framework. Quat. Sci. Rev. 26, 2738e2757.

Schulz, O., Busche, H., Benbouziane, A., 2008. Decadal precipitation variances andreservoir inflow in the semi-arid Upper Draa basin (South-Eastern Morocco). In:Zereini, F., H€otzl, H. (Eds.), Climatic Change and Water Resources in the MiddleEast and North Africa. Springer, Heidelberg, pp. 165e178.

S�ebrier, M., Siame, L., Zouine, E.M., Winter, T., Missenard, Y., Leturmy, P., 2006.Active tectonics in the moroccan high atlas. Compt. Rendus Geosci. 338, 65e79.

St€ablein, G., 1988. Geomorphological aspects of the quaternary evolution of theOuarzazate Basin, Southern Morocco. Lect. Notes Earth Sci. 15, 433e443.

Starkel, L., 2003. Climatically controlled terraces in uplifting mountain areas. Quat.Sci. Rev. 22, 2189e2198.

Stokes, M., Mather, A.E., 2015. Controls on modern tributary-junction alluvial fanoccurrence and morphology: high Atlas Mountains, Morocco. Geomorphology248, 344e362.

Stokes, M., Mather, A.E., Harvey, A.M., 2002. Quantification of River-capture InducedBase-level Changes and Landscape Development, Sorbas Basin, SE Spain, vol.191. Geological Society, London, Special Publications, pp. 23e35.

Stokes, M., Cunha, P.P., Martins, A.A., 2012a. Techniques for analysing Late Cenozoic

M. Stokes et al. / Quaternary Science Reviews 166 (2017) 363e379 379

river terrace sequences. Geomorphology 165, 1e6.Stokes, M., Griffiths, J.S., Mather, A., 2012b. Palaeoflood estimates of Pleistocene

coarse grained river terrace landforms (Río Almanzora, SE Spain). Geo-morphology 149, 11e26.

Stokes, M., Mather, A.E., Belfoul, A., Farik, F., 2008. Active and passive tectoniccontrols for transverse drainage and river gorge development in a collisionalmountain belt (Dad�es Gorges, High Atlas Mountains, Morocco). Geomorphology102, 2e20.

Tes�on, E., Teixell, A., 2008. Sequence of thrusting and syntectonic sedimentation inthe eastern Sub-Atlas thrust belt (Dad�es and Mgoun valleys, Morocco). Int. J.Earth Sci. 97, 103e113.

Teixell, A., Arboleya, M.-L., Julivert, M., 2003. Tectonic shortening and topography inthe central High Atlas (Morocco). Tectonics 22, 1e13.

Thiel, C., Buylaert, J.-P., Murray, A., Terhorst, B., Hofer, I., Tsukamoto, S., Frechen, M.,2011. Luminescence dating of the Stratzing loess profile (Austria) - testing thepotential of an elevated temperature post-IR IRSL protocol. Quat. Int. 234,23e31.

Tjallingii, R., Claussen, M., Stuut, J.-B.W., Fohlmeister, J., Jahn, A., Bickert, T., Lamy, F.,R€ohl, U., 2008. Coherent high- and low-latitude control of the northwest Afri-can hydrological balance. Nat. Geosci. 10, 670e675.

Valero-Garc�es, B., Zeroual, E., Kelts, K., 1998. Arid phases in the western Mediter-ranean region during the last glacial cycle reconstructed from lacustrine re-cords. In: Benito, G., Baker, V.R., Gregory, K.J. (Eds.), Palaeohydrology andEnvironmental Change. John Wiley and Sons Ltd., pp. 67e80

Vandenberghe, J., 2015. River terraces as a response to climatic forcing: formationprocesses, sedimentary characteristics and sites for human occupation. Quat.Int. 370, 3e11.

Vandenberghe, J., Maddy, D., 2001. The response of river systems to climate change.Quat. Int. 79, 1e3.

Wang, X., Van Balen, R., Yi, S., Vandenberghe, J., Lu, H., 2014. Differential tectonicmovements in the confluence area of the Huang Shui and Huang He rivers(Yellow River), NE Tibetan Plateau, as inferred from fluvial terrace positions.Boreas 43, 469e484.

Warme, J.E., 1988. Jurassic carbonate facies of the central and Eastern High Atlas rift,Morocco. Lect. Notes Earth Sci. 15, 169e199.

Weisrock, A., Wengler, L., Mathieu, J., Ouammou, A., Fontugne, M., Mercier, N.,Reyss, J.-L., Valladas, H., Guery, P., 2006. Upper Pleistocene comparative OSL, U/Th and 14C datings of sedimentary sequences and correlative morphodynam-ical implications in the South-Western Anti-Atlas (Oued Noun, 29� N, Morocco).Quaternaire 17, 45e59.

Weissel, J.K., Seidl, M.A., 1997. Influence of rock strength properties on escarpmentretreat across passive continental margins. Geology 25, 631e634.

Westaway, R., Aït Hssaine, A., Demir, T., Beck, A., 2009. Field reconnaissance of theAnti-Atlas coastline, Morocco: fluvial and marine evidence for Late Cenozoicuplift. Glob. Planet. Change 68, 297e310.

Whipple, K.X., Hancock, G.S., Anderson, R.S., 2000. River incision into bedrock:mechanics and relative efficacy of plucking, abrasion and cavitation. Geol. Soc.Am. Bull. 112, 490e503.

Wiche, K., 1953. Klimamorphologische und talgeschichtliche Studien im M'Goun-gebiet. In: Mitt. Der Geographischen Ges. Wien, vol. 95, pp. 4e41.

Wintle, A.G., Murray, A.S., 2006. A review of quartz optically stimulated lumines-cence characteristics and their relevance in single-aliquot regeneration datingprotocols. Radiat. Meas. 41, 369e391.